Systemic Financial Risk of Stock Market Based on Multiscale Networks

This paper investigated the multiscale features of risk spillover networks among financial institutions in stock market at various time-frequency horizons. We propose multiscale risk spillover networks by combining MEMD and generalized vector autoregressive methods. Using daily returns of 36 Chinese publicly listed financial institutions during 2010-2022, we construct multilayer risk spillover networks at short-, medium-, and long-term scale. Moreover, we analyze the topological statistics of multiscale risk spillover networks at the system, sector and institution levels, especially different layers’ similarity and uniqueness. The results indicate that (1) at a system level, the topological structures and evolution behaviors of networks vary across time scales and multiscale networks provide more valuable information on the connectedness among financial institutions. Moreover, some unique features, which could not be detected in a particular single layer, are found in multiscale risk spillover networks, (2) at a sector level, the banks sector is almost the main risk spillover sector in high, medium and low frequency layer, and (3) at an institution level, financial institutions play various roles in receiving or sending shocks based on different risk spillover and contagion channels, and the center of the risk spillover network varies across different time scales. Our multiscale risk spillover networks provide a new perspective for participants in financial market, especially for investors or hedgers who have different investment horizons